Bug 10304

Created attachment 9149 [details]
screenshot of broken kdm screen

Created attachment 9150 [details]
used xorg.conf

Created attachment 9151 [details]
X log

Created attachment 9152 [details]
running xrandr serveral times, one run after another shows different results

OK, we can get that fixed.  Can you provide a physical measurement of your panel size (in mm)?

DisplaySize	  330   210	# mm

these are real values measured manually :)

Still doesn’t work witch xserver (1.2.99.903) and intel driver (1.9.93)

Still doesn’t work with xserver (1.2.99.903) and intel driver (1.9.94)

Unfortunately the problem persist with xserver 1.3.0 and intel driver 2.0.0. Is this issue really so hard? :-)

*** Bug 10756 has been marked as a duplicate of this bug. ***

Created attachment 11068 [details] [review]
proposed fix for physical size detection

Could you test the attached patch against a current server?

Unfortunately the problem persist with xserver 1.3.0 and intel driver 2.1.1 with your patch

Eric to respond...

Created attachment 11949 [details] [review]
patch to ry

This patch prefers the user configured DisplaySize to the probed DDC data, please add the actual physical size to your xorg.conf monitor section and have a try.

(In reply to comment #14)
> Created an attachment (id=11949) [details]
> patch to ry
> 
> This patch prefers the user configured DisplaySize to the probed DDC data,
> please add the actual physical size to your xorg.conf monitor section and have
> a try.
> 

This patch needs probably go upstream anyway (should go in 1.4.1 as well) as there are often cases when we don't know the physical size of an output and need to specify it.  See bug 12474 for example.

Created attachment 11987 [details]
Xorg.0.log

Same problem here on Acer 5022 with Mobility Radeon X700, xserver 1.3.0.0 and latest xf86-video-ati from git

*** Bug 12784 has been marked as a duplicate of this bug. ***

The problem is not caused by the driver, but by bad EDID data from the panel. Bytes 66-67 (image size in mm) are simply copied over from bytes 21-22 (panel size in cm) which is obviously incorrect.

I believe I've fixed the EDID issue upstream now (confused your issue with a very similar one).  Please retest.

*** Bug 12581 has been marked as a duplicate of this bug. ***

Created attachment 12011 [details]
Xorg.0.log

The problem still persists, even with the patch from http://gitweb.freedesktop.org/?p=xorg/xserver.git;a=commit;h=fc092334ac0a323b80a9602cb8bf60ca9dee3bfa

In Xorg.0.log the EDID quirk is mentioned, but xdpyinfo still shows wrong dpi (and fonts are huge)

Created attachment 12012 [details]
xdpyinfo

xdpyinfo output

could anybody post a test result of patch in comment# 14?

(In reply to comment #23)
> could anybody post a test result of patch in comment# 14?
> 

Yes, i patch this patcg again Archlinux xorg-server 1.4 :
  

# Patches from redhat:
  patch -Np3 -i ${startdir}/src/xorg-redhat-die-ugly-pattern-die-die-die.patch 
  patch -Np0 -i ${startdir}/src/xorg-x11-server-0.99.3-init-origins-fix.patch 
  patch -Np1 -i ${startdir}/src/xorg-x11-server-1.0.1-Red-Hat-extramodes.patch 
  patch -Np0 -i ${startdir}/src/xorg-x11-server-1.1.0-no-move-damage.patch 
  patch -Np1 -i ${startdir}/src/xorg-x11-server-1.1.1-vbe-filter-less.patch 
  patch -Np1 -i ${startdir}/src/xorg-x11-server-1.1.1-no-scanpci.patch 
  patch -Np1 -i ${startdir}/src/xserver-1.3.0-newglx-offscreen-pixmaps.patch 

  # Own patches
  patch -Np1 -i ${startdir}/src/xorg-server-1.2.0-defaultdpi.patch 
  patch -Np1 -i ${startdir}/src/prefer-displaysize.patch
  
  aclocal
  autoconf
  automake --add-missing
  ./configure --prefix=/usr \
              --with-mesa-source=${startdir}/src/Mesa-${_mesaver} \
              --enable-ipv6 \
              --enable-dri \
              --enable-dmx \
              --enable-xvfb \
              --enable-xnest \
              --enable-composite \
              --enable-xcsecurity \
              --enable-xorg \
              --enable-xtrap \
              --enable-xevie \
	      --enable-xephyr \
              --enable-glx-tls \
	      --disable-kdrive \
              --enable-install-setuid \
              --disable-xprint \
              --sysconfdir=/etc/X11 \
              --localstatedir=/var \
              --with-default-font-path=/usr/share/fonts/misc,/usr/share/fonts/100dpi:unscaled,/usr/share/fonts/75dpi:unscaled,/usr/share/fonts/TTF,/usr/share/fonts/Type1 \
              --with-rgb-path=/usr/share/X11/rgb \
              --with-xkb-path=/usr/share/X11/xkb \
              --with-xkb-output=/usr/share/X11/xkb/compiled \
              --with-dri-driver-path=/usr/lib/xorg/modules/dri \
              --disable-static

  make || return 1
  make DESTDIR=${startdir}/pkg install || return 1
  mkdir -p ${startdir}/pkg/etc/X11
  mkdir -p ${startdir}/pkg/var/lib
  mv ${startdir}/pkg/usr/share/X11/xkb/compiled ${startdir}/pkg/var/lib/xkb
  ln -sf /var/lib/xkb ${startdir}/pkg/usr/share/X11/xkb/compiled

  cd ${startdir}/pkg/usr/lib/xorg/modules/extensions
  mv libGLcore.so libGLcore.xorg
  mv libglx.so libglx.xorg

  # prevent conflict with nvidia-utils
  mv ${startdir}/pkg/usr/lib/xorg/modules/libwfb.so \
     ${startdir}/pkg/usr/lib/xorg/modules/libwfb.so.1.4
-------------------------------------------------------------

ARCH x86_64

kernel 2.6.23
dbus 1.1.2
hal 0.5.10
xf86-video-intel 2.1.1 
xorg 1.4
GCC 4.2.2
pixman 0.9.5
-----------------------------------------------------------------

Result:  X server not starting, crash!

Backtrace:
0: X(xf86SigHandler+0x6a) [0x48d0ba]
1: /lib/libc.so.6 [0x2b8cc69517b0]
2: X(xf86RandR12CreateScreenResources+0x1ee) [0x4b2a5e]
3: X [0x4ada9f]
4: X(main+0x319) [0x4367d9]
5: /lib/libc.so.6(__libc_start_main+0xf4) [0x2b8cc693eae4]
6: X(FontFileCompleteXLFD+0x261) [0x435c79]

Fatal server error:
Caught signal 11.  Server aborting


please see Xorg.0.log - DPI is not anyway (96,96) , but (96,154) 

Created attachment 12171 [details]
crashed xorg-server 1.4

Created attachment 12172 [details]
crashed xorg-server 1.4 log

crashed xorg-server 1.4 log

Confirming.

We tested #14 when it appeared but the server crashed. Unfortunately we had no debug symbols compiled in so we weren't able to dig deeper what's the cause. Now I see that the patch was problematic as crash happens for others, too.

Looks like you're not using a monitor section at all or do not specify it properly with 

  Option "Monitor-outputname" "monitorsection"

Still it's a bug in the patch of comment #14, since you get a Nullpointer
reference when not using a monitor section. This should not happen. I stumbled
across this today as well ...

Better version of this patch:

diff --git a/hw/xfree86/modes/xf86RandR12.c b/hw/xfree86/modes/xf86RandR12.c
index 38435c9..135356d 100644
--- a/hw/xfree86/modes/xf86RandR12.c
+++ b/hw/xfree86/modes/xf86RandR12.c
@@ -426,8 +426,18 @@ xf86RandR12CreateScreenResources (ScreenPtr pScreen)
            xf86OutputPtr   output = config->output[config->compat_output];
            xf86CrtcPtr     crtc = output->crtc;
 
-           if (crtc && crtc->mode.HDisplay &&
-               output->mm_width && output->mm_height)
+           if (output->conf_monitor &&
+               (output->conf_monitor->mon_width  > 0 ||
+                output->conf_monitor->mon_height > 0))
+           {
+               /*
+                * Prefer user configured DisplaySize
+                */
+               mmWidth = output->conf_monitor->mon_width;
+               mmHeight = output->conf_monitor->mon_height;
+           }
+           else if (crtc && crtc->mode.HDisplay &&
+                    output->mm_width && output->mm_height)
            {
                /*
                 * If the output has a mode and a declared size, use that

I committed a slightly modified version of this patch (a monitor section with either width or height <= 0 shouldn't be honored as well).
But I think the original issue is still open.

(In reply to comment #29)

it's a good thing, thanks! Xserver not crashed. 

-------------------------------------------------------------------------------
I use new add (for my laptop) in xorg.conf for  Option "Monitor-outputname" "monitorsection" :

Section "Monitor"
        Identifier      "LCD 1280x800"
        Option          "DPMS"          "true"
	VendorName	"LPL"
	DisplaySize     338 211
	Option 		"Monitor-LVDS " "LVDS  monitor"
EndSection
-------------------------------------------------------------------------------

Results : xserver 1.4 with "better" patch set display dimensions now right :
(**) intel(0): Display dimensions: (338, 211) mm   

but DPI is set wrong - DPI is now 96,154.

Created attachment 12180 [details]
better patch - xserver works ( Xorg.0.log)

(In reply to comment #30)
> 
> Results : xserver 1.4 with "better" patch set display dimensions now right :
> (**) intel(0): Display dimensions: (338, 211) mm   
> 
> but DPI is set wrong - DPI is now 96,154.
> 

check xdpyinfo.  you dpi should be correct. the (**) intel(0): DPI set to (96, 154) line is misleading in that it takes into account the virtual size of the desktop when it prints that.  The xrandr code does the right thing however.

(In reply to comment #32)

OK, results from xdpyinfo:

# xdpyinfo |grep resolution
 resolution:    112x968 dots per inch
 
the font is huge

---------------------------------------------------------

when i use  # xrandr --fbmm 338x211 : (338x211 is physical size of my laptop´s display)


# xdpyinfo |grep resolution
  resolution:    96x96 dots per inch
the font is correct

>Section "Monitor"
>        Identifier      "LCD 1280x800"
>        Option          "DPMS"          "true"
>        VendorName      "LPL"
>        DisplaySize     338 211
>        Option          "Monitor-LVDS " "LVDS  monitor"
>EndSection
Isn't the monitor Option line just wrong? Shouldn't this be

  Option          "monitor-LVDS " "LCD 1280x800"

instead?

(In reply to comment #34)
> >Section "Monitor"
> >        Identifier      "LCD 1280x800"
> >        Option          "DPMS"          "true"
> >        VendorName      "LPL"
> >        DisplaySize     338 211
> >        Option          "Monitor-LVDS " "LVDS  monitor"
> >EndSection
> Isn't the monitor Option line just wrong? Shouldn't this be
> 
>   Option          "monitor-LVDS " "LCD 1280x800"
> 
> instead?
> 


Yes, right !! I am idiot, sorry.

If i use "better" patch + in xorg.conf DisplaySize + Option "Monitor-outputname" "monitorsection" the dpi is correct !! 96dpi, no huge wrong DPI.

The problem is fixed.

Can another make test for confirmation?

Created attachment 12190 [details]
xorg.conf with right set DisplaySize and "Monitor-outputname" "monitorsection"

Created attachment 12191 [details]
xdpyinfo with fixed DPI

Since Matthias pushed the fix upstream, let's finally close this one as fixed.

Since Matthias pushed the fix upstream, let's finally close this one as fixed.

Thanks to all !

Here on Ubuntu we have the bug:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/107320

It seems we have someone with a similar but not identical monitor.

servaas-xubuntu-user, bug #127827
(II) intel(0): Manufacturer: LPL Model: 2a00 Serial#: 0 (near miss!)
(II) intel(0): Setting screen physical size to 289 x 21
servaas-xubuntu-user (Problem in XUbuntu in general, not just GDM)
X Window System Version 1.3.0
X Protocol Version 11, Revision 0, Release 1.3
Current Operating System: Linux ubuntu 2.6.22-8-generic #1 SMP Thu Jul 12 15:59:45 GMT 2007 i686
(==) intel(0): DPI set to (75, 75)

His bug is reported in Ubuntu at:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/127827
and it's Xorg.0.log file is at:
http://launchpadlibrarian.net/8675708/Xorg.0.log

My believe is that the patch should be modified to add this monitor model (2a00).

We also have 1 person with detected 40x30 mm screen, and one with 10mmx10mm screen. I don't think they are related to this bug, but if curious more info is at: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/151311/comments/29

*** Bug 13458 has been marked as a duplicate of this bug. ***

Alois, can you please add Option "ModeDebug" "true" to your xorg.conf and attach the new log? Apparently there are several panels from Philips with product ID = 0 and not all of them need this quirk. I am wondering if there is some more identification we can use.

(In reply to comment #43)
> Alois, can you please add Option "ModeDebug" "true" to your xorg.conf and
> attach the new log? Apparently there are several panels from Philips with
> product ID = 0 and not all of them need this quirk. I am wondering if there is
> some more identification we can use.
> 

	
yes, but now using xorg-server 1.6 and xf86-video-intel 2.6.99.902

will have any influence?

Created attachment 24444 [details]
ModeDebug On

Reopening as there seems to be other models affected.  I'll attach the Xorg.0.log with ModeDebug on for my system as well.

Created attachment 25646 [details]
Xorg log

xorg-x11-drv-ati-6.12.2-11.fc11.x86_64

HP Pavilion zv6000

This bug report is way too big.  The original bug should be fixed, and the reporter no longer has the hardware.  If people have bugs with other monitors, please open your own bugs so you actually get some attention!